TREM-1 ligand expression on platelets enhances neutrophil activation

Sepsis is a complex clinical condition caused by a dysregulated immune response to an infection oftentimes resulting in multiorgan failure with a fatal outcome.¹  Current concepts for the understanding of this dramatic clinical condition suggest that an overwhelming innate inflammatory response to a microbial infection associated with the release of factors such as IL-1, IL-6, TNF-α, and others early on is involved.² 

Recently, the triggering receptor expressed on myeloid cells 1 (TREM-1) has been identified as an important receptor involved in the innate inflammatory response and in sepsis.^3-6  TREM-1 is a member of the v-type immunoglobulin super family and expressed on polymorphonuclear leukocytes (PMNs) and monocytes.⁷  The expression of TREM-1 is up-regulated upon stimulation with microbial products, and receptor ligation activates the full repertoire of PMN effector functions such as the respiratory burst, phagocytosis, release of IL-8, and myeloperoxidase in synergy with Toll-like receptor (TLR) ligands such as LPS or bacterial lipopeptides.^3,6,8,9  The importance of TREM-1 in the innate inflammatory response is underlined in mouse models for sepsis where the administration of a recombinant soluble TREM-1 fusion protein or a conserved TREM-1 peptide can save the animals from a lethal endotoxic shock or microbial sepsis induced by cecal ligation and puncture.^5,6  The clinical significance of the TREM-1 system is further emphasized by reports that a soluble form of TREM-1 (sTREM-1) is released and detectable in the bronchoalveolar lavage fluid or serum in patients with ventilator-associated pneumonia or sepsis, respectively.^10-12  Among the critically ill patients analyzed in these studies, the determination of sTREM-1 proved to be a useful and highly sensitive parameter for accurate diagnosis. Further studies where sTREM-1 is detected in patients with chronic obstructive pulmonary disease,¹³  peptic ulcer disease,¹⁴  or inflammatory bowel disease¹⁵  confirm the involvement of TREM-1 also in noncritically ill patient collectives and suggest that the activation of TREM-1 plays a general role in the innate inflammatory response.⁷  However, a deeper understanding of how TREM-1 influences inflammatory responses and sepsis requires the characterization of the natural ligand for TREM-1 and its expression pattern.⁷ 

To this end, we provide evidence that the natural ligand for TREM-1 is present on human platelets. We demonstrate specific binding of recombinant soluble TREM-1 (rsTREM-1) on human platelets. In addition, we show that coincubation of PMNs with platelets in the presence of microbial LPS enhances the neutrophil respiratory burst and release of IL-8 as primary PMN effector functions in a TREM-1–specific manner. Taken together, our results indicate a yet-unrecognized interaction between PMNs and platelets during innate the inflammatory response that is mediated by TREM-1 and its ligand.

Lipopolysaccharide (LPS) from Salmonella typhimurium was obtained from Sigma-Aldrich (Taufkirchen, Germany). The recombinant sTREM-1 fusion protein was generated as described in the next paragraph and purified as described previously.¹⁶  Human IgG was purified by protein A affinity chromatography (Millipore, Volketswil, Switzerland) from serum of healthy volunteers. LP17 spanning the complementary determining region loop 3 of TREM-1 (LQVTDSGLYRCVIYHPP) and a scrambled control peptide (TDSRCVIGLYHPPLQVY) were chemically synthesized as a COOH terminally amidated peptide (Pepscan Systems, Lelystad the Netherlands). All proteins were endotoxin free (< 0.1 EU/μg protein) determined by a limulus amebocyte lysate assay (QCL-1000; BioWhittaker, Verviers, Belgium). The following mAbs were used for fluorescence-activated cell sorting (FACS) analysis: FITC-labeled mAbs against CD66b (Coulter Immunotech, Hamburg, Germany), PE-labeled mAbs against CD62P (eBioscience, San Diego CA), PE-Cy7 labeled mAbs against CD45, allophycocyanin (APC)–labeled mAbs against CD41a (both from BD Pharmingen, Heidelberg, Germany), APC-Cy7–labeled mAbs against CD62L (Caltag, Hamburg, Germany), and respective isotype controls (all purchased from BD Pharmingen). Other antibodies used were mAbs against CD18 clone IB4 (Ancell, Bayport, MN), PSGL-1 clone PL-1 (Bender Medsystems, Vienna, Austria), TREM-1 clone 6B1, raised by fusion of SP2/0 myeloma cells (from American Type Culture Collection, Manassas, VA) with splenocytes from a BALB/c mouse immunized with a recombinant sTREM-1 fusion protein and screened against TREM-1, and monoclonal mouse IgG1 clone 4C9 (previously described¹⁶ ). All human studies were performed after obtaining informed consent from healthy volunteer donors in accordance with the Declaration of Helsinki and were approved by the Landesaerztekammer Rhineland-Palatine Ethics Committee according to the institutional guidelines.

For the production of a soluble form of TREM-1 (accession no., BC017773) the cDNA encoding for the extracellular domain (aa 21-200) after prediction of signal-peptide cleavage site with SPScan and Signalseq from the GCG package (version 10.2; Accelrys, Cambridge, United Kingdom) was amplified from cDNA clone (RZPD clone ID no. IMAGp958N091513Q2 from the German Resource Center Data base RZPD, Berlin, Germany) using primer adding 5′ and 3′ cleavage sites for AatII and BsmBI, respectively. This allowed an in-frame integration into a eukaryotic expression vector coding for the leader sequence, hinge, CH2, and CH3 region of a human heavy chain of an Igγ1 molecule. The amino acid cysteine at position 220 (Kabat nomenclature) in the hinge region forming a disulfide bond to the light chain of an antibody was substituted against serine. The expression of the 92.9-kDa fusion protein was controlled by a 1.2-kB promoter and 1.3-kB 3′UTR of the murine immunoglobulin heavy-chain locus. After linearization with AhdI, the plasmid (based on pcDNA3) was transfected by electroporation (230 V; 975 μF) into mouse myeloma cell line Sp2/0. Stable transfectants were selected with 1 mg/mL G418.l

Recombinant sTREM-1 or Hu IgG in 0.1M NaHCO₃ and biotin (Sigma-Aldrich) in DMSO were incubated for 2 hours at RT in ratio of 10:1, and then centrifuged. Nonbinding biotin was removed with a PD10 desalting column (Amersham, Freiburg, Germany).

For platelet isolation, citrated blood (11 mM sodium citrate from Sigma-Aldrich) was centrifuged at 100g for 15 minutes at room temperature to obtain platelet-rich plasma (PRP). PRP was layered on 34% (wt/vol) BSA (Sigma-Aldrich) and centrifuged at 550g for 10 minutes. Purified platelets were collected from the interphase and washed twice with human Tyrode buffer + 5 mM EGTA (HuTyrodeB: 5 mM Hepes, 137 mM NaCl, 2.7 mM KCl, 11.9 mM NaHCO3, 1 mM MgCl, 0.1% BSA, 1% glucose). For activation, platelets were incubated for 10 minutes with 5 U/mL thrombin (Sigma-Aldrich) at 37°C and subsequently fixed with 2% (wt/vol) paraformaldehyde (PFA; Sigma-Aldrich). Residual PFA was removed by two additional washing steps. Contamination by other blood cells (mainly erythrocytes) in platelet suspensions was always lower than 0.5%.

PMNs were separated from heparinized/citrated blood using Polymorphprep (Nycomed, Oslo, Norway) as described previously.¹⁶  Purity of cell preparation was assessed by FACS using CD66b as the marker for PMNs. In general, 92% to 98% of cells were CD66b⁺ PMNs.

Platelet aggregation was measured with an optical aggregometer (Apact 4s plus; Rolf Greiner Biochemica, Flacht, Germany) and corresponding software. Platelet-rich plasma (200 μL) and indicated stimuli were measured for more than 500 seconds for their aggregation state.

Isolated stimulated or unstimulated platelets were stained with biotinylated recombinant sTREM-1 or a biotinylated control human IgG in concentrations as indicated for 20 minutes, washed twice, and stained with PE-conjugated streptavidin.

Citrated blood was immediately mixed with the same volume of 37°C PBS. Of the blood samples, 100 μL was placed into (12 × 75 mm) polystyrene tubes (Falcon; Becton Dickinson, Heidelberg, Germany), stimulated as indicated, and cultured at 37°C, 5% CO₂ in air. Staining with mAbs began 15 minutes before stimulation was stopped by addition of 2 mL FACS lysing solution (Becton Dickinson). Samples were incubated at room temperature for 10 minutes and washed twice with FACS buffer and analyzed by flow cytometry using a FACScanto and Diva software (Becton Dickinson).

Isolated platelets were incubated for 30 minutes with 0.5 μg/mL calcein (InvitroGen, Heidelberg, Germany) or 4 μM PKH26 (Sigma-Aldrich) in medium containing 3% FCS, then washed twice. After addition of platelets into FACS tubes containing PMNs (1:100 volume), samples were immediately stimulated and measured with FACS analyzer at different time points.

The presence of hydrogen peroxide was detected by oxidation of dichloro-fluorescein diacetate (DCFH-DA from Sigma-Aldrich) into green fluorescent DCF. For the kinetic analysis, cells were kept in a fluorescence reader (SpectraFluor 4; Tecan, Crailsheim, Germany) as described previously.¹⁶  Specific fluorescence index (SFI) of stimulated cells was obtained by subtraction of the background fluorescence of labeled cells incubated in medium alone at the corresponding time points.

For analysis by enzyme-linked immunosorbent assay (ELISA), supernatants were derived from stimulation with purified PMNs and fixed platelets after 8 hours and frozen at −20°C until required. Supernatants were analyzed by ELISA for IL-8 (R&D Systems, Wiesbaden, Germany) according to the manufacturer's instruction.

Statistical analyses of the data were performed by Student t test. P values less than .01 were considered statistically significant.

Hypothesizing that the ligand for TREM-1 is of endogenous origin⁷  and of importance in the immediate innate immune response during infection, we screened hematopoietic cells for specific binding of a recombinant soluble fusion protein consisting of the extracellular domain of human TREM-1 and the Fc part of human IgG1 (rsTREM-1). We were unable to detect binding of rsTREM-1 on PMNs, monocytes, or lymphocytes (Figure 1A). Since platelets are known to be involved in inflammatory processes such as sepsis,^17,18  we decided to also study platelets for rsTREM-1 binding. As shown in Figure 1, we noted that human platelets were able to bind biotinylated rsTREM-1 (Figure 1B) in a concentration-dependent manner on unstimulated or thrombin-activated platelets (Figure 1C). The specificity of this binding was further indicated by reduced binding of labeled rsTREM-1 in the presence of excess unlabeled rsTREM-1, but not in the presence of an irrelevant protein (Figure 1B,D). The binding of rsTREM-1 on platelets was also reduced in the presence of a conserved TREM-1–specific peptide called LP17 that has previously been demonstrated to be protective in LPS-induced sepsis or pneumonia,^5,19  but not in the presence of a scrambled control peptide (Figure 1E).

These results suggest that a ligand for TREM-1 is present on human platelets and that the expression of this ligand is not up-regulated upon activation. This is of special interest as the interaction between PMNs and platelets, especially under inflammatory conditions, has been described previously.²⁰ 

To address whether the interaction of TREM-1 with the putative ligand induces platelet activation, we incubated rsTREM-1 with purified platelets and analyzed platelet activation by monitoring aggregation and degranulation. As depicted in Figure 2, rsTREM-1 did not induce any activation in terms of aggregation (Figure 2A) or degranulation (Figure 2B) as measured by the surface up-regulation of P-selectin (CD62P). To additionally investigate whether cross-linking of the putative ligand is necessary for activation, biotinylated rsTREM-1 was again incubated with platelets and streptavidin was added to mimic multimerization of the receptor. Cross-linking did not induce platelet aggregation or degranulation (data not shown).

These results suggest that the putative ligand for TREM-1 on platelets does not mediate activation of platelets (ie, by reverse signaling).

We and others have previously reported that ligation of TREM-1 using agonistic mAbs induces the activation of PMN effector functions such as the respiratory burst or IL-8 release in synergy with TLR ligands.^3,8,9 

To investigate whether TREM-1 ligand expression on platelets has a functional impact on the initiation of neutrophil effector functions, we incubated platelets with PMNs in the presence or absence of LPS and assessed the respiratory burst activity. To avoid influences by soluble factors released by platelets such as lipid mediators (ie, leukotrienes or PAF), the platelets were fixed with paraformaldehyde before adding them to PMNs. As depicted in Figure 3, LPS alone induced the respiratory burst in a concentration-dependent manner (black filled bars), while platelets alone added to PMNs did not result in any activation under these conditions. However, the presence of platelets greatly enhanced the LPS-initiated respiratory burst (Figure 3 open bars), very much resembling the respiratory burst activity achieved when using TREM-1–specific mAbs for receptor ligation as reported previously.^8,9  When platelets preactivated with thrombin (Figure 3 gray filled bars) were added to PMNs, the enhancement of the respiratory burst induced by low amounts of LPS was even more pronounced compared with nonactivated platelets (Figure 3).

To directly address whether the observed enhancement of LPS-induced PMN activation in the presence of platelets is dependent on TREM-1/ligand interaction, we coincubated PMNs and platelets with LPS in the presence of the rsTREM-1 or a TREM-1–specific mAb and analyzed the induction of the respiratory burst or release of IL-8. As shown in Figure 4, the respiratory burst activity or release of IL-8, respectively, was suppressed only in presence of TREM-1–specific mAb (A,B) or rsTREM-1 (C,D), but not in the presence of the control protein or mAb, indicating that the initiation of the respiratory burst is dependent on the TREM-1/ligand interaction.

Previous reports demonstrate that leukocytes associate with platelets, especially under inflammatory conditions.^17,18,20,21  To visualize this interaction, we incubated platelets with PMNs in the presence or absence of LPS and analyzed conjugate formation by flow cytometry.

As shown in Figure 5, we found that platelets associated readily with PMNs or monocytes, but not with lymphocytes (Figure 5A). Conjugate formation of PMNs and monocytes, but not of lymphocytes, was significantly enhanced in the presence of LPS (Figure 5A) or when thrombin-activated platelets were used (Figure 5B). However, this association between PMNs and platelets could not be inhibited by recombinant sTREM-1 or by TREM-1–specific mAbs (Figure 5C), indicating that interactions other than TREM-1 with its ligand are important to stabilize PMN-platelet conjugates. To further characterize the PMN/platelet interaction, we concurrently analyzed the induction of the respiratory burst and conjugate formation using PKH26-labeled platelets and DCFH-DA–loaded PMNs in the presence or absence of LPS. As shown in Figure 4, we found that the respiratory burst activity was increased in the presence of platelets (Figure S1, available on the Blood website; see the Supplemental Materials link at the top of the online article). Interestingly, this enhanced burst was detectable not only in PMNs in direct contact with platelets, but also in PMNs not associated with platelets, suggesting that conjugate formation is either reversible or mediated by transactivation by a soluble mediator. However, the observation that this enhancement is also detectable in experiments with fixed platelets (Figures 3, 4) indicates that this potential factor is not platelet derived.

The interaction via selectins (PSGL-1) and integrins (Mac-1) is well documented to be important for PMN/platelet association.^22-24  Therefore, we analyzed conjugate formation in the presence of inhibitory mAbs against PSGL-1 or CD18. As demonstrated before by others, we found that both molecules are essential for the interaction of PMNs with resting (Figure 5D filled bars) or activated (Figure 5D open bars) platelets. These results confirm that selectin as well as β₂-integrin–dependent interactions are required for the physical PMN-platelet communication, while the TREM-1/ligand contact is dispensable for this association.

Consequently, inhibition of selectin- or integrin-dependent PMN/platelet interaction with mAbs abrogated the platelet-induced effects on the LPS-mediated respiratory burst (Figure 6). These results support the view that the platelet-PMN interaction is stabilized and enhanced by selectins/integrins and also important for the induction of PMN effector functions.

It is common knowledge that patients with severe infections or sepsis may present with low platelet counts due to activation and sequestration in the periphery.^17,18,21,25 

In sepsis, it has been demonstrated that platelets show enhanced adherence to leukocytes¹⁷  and are activated,¹⁸  although the severity only of multiorgan failure, but not of sepsis, correlates with irreversible platelet activation.²⁶ 

Under experimental flow conditions, platelets seem to interact preferentially with monocytes rather than PMNs, which are dependent on divalent cations and the expression of PSGL-1 and in part also on β1- and β2-integrins for fMLP-activated monocytes.^27,28  However, after severe trauma or sepsis, platelets associate significantly with both monocytes as well as PMNs.^17,29  Recent data from a mouse model for acid-induced acute lung injury (ALI) suggest that the interaction between PMNs and platelets is crucial for the development of ALI also in LPS/zymosan-induced sepsis,³⁰  underlining the importance of the PMN-platelet interaction in vivo. So far, this contact and the associated leukocyte activation have been attributed to the interaction of selectins and integrins expressed by platelets and PMNs or monocytes, respectively.^21-24,30  By detecting a ligand for TREM-1 on platelets, we provide the first evidence for a so-far unrecognized direct link for this specific TREM-1 interaction between PMNs and platelets to the regulation of innate inflammatory responses. Further studies addressing the significance of the TREM-1/ligand interaction in the crosstalk of monocytes with platelets are ongoing.

TREM-like transcript 1 (TLT-1) has been described as a novel receptor in the TREM gene cluster expressed on platelets and megakaryocytes.⁷  In contrast to TREM-1, the cytoplasmic tail of TLT-1 harbors an inhibitory ITIM motif and therefore is predicted as an inhibitory receptor on platelets. TLT-1 colocalizes with CD62-P in α-granules in resting platelets and both receptors become rapidly up-regulated upon activation (ie, with thrombin).³¹  Interestingly, TLT-1 ligation enhances Ca influx rather than inhibiting it, suggesting activatory functions of this receptor.³²  In addition, a conserved TREM-1–specific peptide LP17 that inhibits binding of rsTREM-1 to platelets and ameliorates the effects of septic shock in vivo is predicted to also bind to TLT-1.^5,33  However, in contrast to TLT-1, TREM-1 ligand on platelets is not up-regulated upon activation with thrombin and does not induce platelet activation. Furthermore, the binding of rsTREM-1 is not reduced in the presence of α-TLT-1 mAbs (data not shown), indicating that TLT-1 is not the ligand for TREM-1.

Recent results by Mohamadzadeh et al also describe a function for TREM-1 in filovirus infection.³⁴  The authors demonstrate that Marburg or Ebola virus activates PMNs in a TREM-1–dependent manner, suggesting that a ligand for TREM-1 is present in filoviruses. However, previous data from murine models for sepsis suggest the additional presence of an endogenous ligand for TREM-1.⁷  Here we demonstrate that this endogenous ligand for TREM-1 is expressed by human platelets and functionally relevant for the TLR ligand-induced activation of PMNs. Interestingly, the TREM-1/ligand interaction is not involved in the physical association of platelets with PMNs that is mediated by β2-integrins (ie, CD18) and selectins (PSGL-1) as demonstrated before.^22,24  Previous reports find that contact-dependent PMN activation by platelets is dependent on Mac-1.^22-24  However, in our experimental setup, we used fixed platelets in the absence of serum and did not detect PMN activation in the absence of LPS, suggesting that for the initiation of the respiratory burst in our setting TREM-1–specific signals are required (Figure 4). These integrin/selectin-mediated PMN-platelet interactions are apparently crucial for TREM-1/ligand-induced PMN activation since the respiratory burst in this situation is abrogated in the presence of specific mAbs. This is further emphasized by our observation of enhanced conjugate formation and activation of PMNs by thrombin-activated platelets, albeit the TREM-1 ligand on platelet is not up-regulated upon activation with thrombin. This may be explained by an integrin/selectin-mediated prolongation of the crosstalk between PMNs and platelets.

Inhibition of the LPS-induced respiratory burst and release of IL-8 by rsTREM-1 or by a TREM-1–specific mAb indicate that soluble TREM-1 detected in septic patients might indeed be an anti-inflammatory decoy receptor as suggested previously by Gibot and Massin.³⁵  Furthermore, the presence of a TREM-1 ligand on platelets could explain the previous observation that a ligand for TREM-1 is present on septic patients,³⁶  which might be due PMN-platelet conjugates in the peripheral blood, frequently observed in these patients.¹⁷ 

So far, our attempts to identify this TREM-1 ligand from platelet lysates by affinity chromatography and mass spectrometry were unsuccessful, possibly due to a relatively low affinity of recombinant sTREM-1 to its ligand. More advanced purification strategies using additional cross-linking are under way and, it is hoped, will disclose it.

Taken together, we provide evidence for the presence of a ligand for TREM-1 on human platelets that mediate neutrophil activation in the presence of microbial products such as LPS. By localizing the TREM-1 ligand on platelets, we can shed new light on the TREM-1–mediated innate inflammatory response that might contribute to future concepts in the treatment of overwhelming immune responses in the course of sepsis.

This work was supported by grants from the Deutsche Forschungsgemeinschaft (Ra 988/2–1/2 and Ra 988/3–1 to H.S. and M.P.R.).

The authors thank Andrea Drescher and Ann-Katrin Meinl for excellent technical assistance. We thank Dr Stefan Tenzer (Institute for Immunology, Mainz, Germany) for critical review of the paper.

Contribution: P.H. performed the experiments (the work is part of his PhD thesis); L.G.-H. constructed the recombinant sTREM-1 fusion protein; P.L. performed platelet aggregation assays; H.S. contributed to the design of the project and wrote the paper; M.P.R. designed the project and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Markus P. Radsak or Hansjörg Schild, Institute for Immunology, University of Mainz, Obere Zahlbacher Str 67, D-55131 Mainz, Germany; e-mail: radsak@uni-mainz.de or schild@uni-mainz.de.